Skip to content

Tous les séminaires

Ce Séminaire est reporté au 10 Avril 2014

Jeudi 19 Décembre 2013 - de 14h à 15h30, Salle S016, INSEE-MK2, 15 Boulevard Gabriel Péri, 92245 MALAKOFF Cedex

(Métro : Malakoff/Plateau de Vanves)

Alfred Galichon (Sciences Po)

"mu-Quantile Regression"

"We propose a version of Quantile Regression, called -Quantile Regression, that (i) allows for a factor representation Y = β(U) X, even in the misspecied case, and (ii) is easily extended to the multivariate case. In the univariate case, µ-Quantile Regression is obtained from classical Quantile Regression by imposing a natural monotonicity requirement in the dual formulation of the latter. -Quantile Regression has a general formulation which embeds the Monge-Kantorovich theorem on optimal transportation. Several applications to diverse problems such as multiple Engel curve estimation, and measurement of nancial risk, are considered."